Question

I have a C header file that I have precompiled using the gcc -E flag, and am now trying to parse using Lex and Yacc; however, it is getting hung up on typedef’d variables.

for example:

typedef unsigned long ULONG;
ULONG i = 5;

will throw a syntax error at the second line’s ULONG.

I have tried to redefine part of the grammar (found here) http://www.quut.com/c/ANSI-C-grammar-y.html, specifically under type_specifer by replacing TYPE_NAME with IDENTIFIER, however that create multiple s/r and r/r errors that I am unable to fix.

Are there other approaches that you would recommend?
Or a different approach at precompiling all together?

Answer 1

In the code you link too:
http://www.quut.com/c/ANSI-C-grammar-y.html,

look at the bottom:
You will see this function

int check_type(void)
{
 /*
  * pseudo code --- this is what it should check
  *
  * if (yytext == type_name)
  *     return TYPE_NAME;
  *
  * return IDENTIFIER;
  */

/*
 *  it actually will only return IDENTIFIER
 */

    return IDENTIFIER;
}

You actually need to write the code that identifies weather a sequence (yytest, yytext+yylength] is an identifier or a TYPE_NAME. This means in the parser you need to build some structure as you parse the code so that this function can look up the identifier in the structure.

Pre-populate the structure with the default types char/int/short/long/float/double etc. The other types you will need to add as you parse the input.